Delay, Loss and Throughput in Packet-Switched Networks
When a package is sent from one host (source) to another host (destination), it travels through a series of nodes and routers. Travelling from one of these nodes/routers and on to a subsequent node/router, the packet will suffer from different types of delay. The most notable and important delays are the nodal processing delay, queuing delay, transmission delay, and propagation delay. All these different delays sums up in a total nodal delay. Types of Delay A packet is sent from the upstream node through router A and then to router B. Let’s have a look at the process and analyze the total no of delay. Processing Delay The time used to examine the packet’s header and determine where to direct it,and check for bit-level errors in the packet that occurred in transmission is called the processing delay. There are also several other factors that affects this delay, for example: The need to check for bit-level errors that has occurred during transmission to the router. Queuing Delay The queuing delay is the time it takes for the packet to be transmitted onto the link. Naturally; the length of this time is defined by the number of packets that was added to the queue prior to this packet. It is also very important that the rate of incoming packets does not exceed the rate of outgoing packets. If this happens, the delay will approach infinity, and the transmission could suffer packet losses. Transmission Delay Packets are commonly transmitted in a first-come-first-serve manner, and it is realistic to assume that a packet will not be transmitted until all prior packets are transmitted. The transmission delay is the amount of time used to transmit (push) all of the packets bits from the queue and into the link. To calculate the delay, the following formula is used: \frac{L}{R} , where L is the length of the packet (in bits), while R is the transmission speed of the Ethernet link in Mbps (Megabits per second). Propagation delay As the packet does not magically appear straight away on router B after it has been pushed into the link from router A, there is of course a propagation delay. This is the time required to propagate from the beginning of the link and to router B. The speed of the propagation heavily depends on the physical medium of the link (fibre optics, twisted pair, copper wire, etc). This delay can be calculated by the formula: \frac{d}{s} , where d is the distance between router A and router B, and s is the propagation speed between the two routers. The speed is usually in the range 2 · 108 meter/sec to 3 · 108 meter/sec, which basically is around the speed of light. Packet Loss in Package-Switched Networks In the queuing delay section earlier, it was mentioned that the delay time will approach infinity; this is of course not realistic as the queues usually has a finite number of packets which can be queued. Basically this means that when sending a lot of packets in to a queue at a high rate (or at the same time), packet loss will be experienced as the queue will be m''axed out'' and the router will drop packets. This will start a “chain-reaction” of increasing the rate of incoming packets, as the dropped packets will need to be retransmitted to the router. A lost packet can be retransmitted on an end-to-end basis in order to ensure that all data are eventually transferred from source to destination. How fast data can be passed through the network media is the throughput of media. Throughput in Computer Networks Imagine a large file being sent from Host A ''to ''Host B across a computer network. The instantaneous throughput at any instant of time is the rate (in bits/sec) at which Host B is receiving the file. The average throughput of the file is \frac{F}{T} bits/sec, where the file consists of F bits and the transfer time is T (in seconds). Category:Alle Category:Fullført